SelectionHapStats is a repository of Python scripts written to identify natural selection events in the genome and R scripts written to visualize the signatures of selective sites. The python code provided calculates haplotype homozygosity statistics H12 and H2/H1 in a genome-wide scan, as well as identified H12 peaks in genomic data. The R code provided visualizes the haplotype frequency spectra for the top peaks in the data and the genome-wide scan of H12.

This code presented in this repository is based on the arXived paper, Recent selective sweeps in North American Drosophila melanogaster show signatures of soft sweeps (http://arxiv.org/abs/1303.0906).

Check out my blog post for further description on the project and examples of visual output from the code!

URL: https://github.com/ngarud/SelectionHapStats

I have been awarded a fellowship from the Stanford Center for Human Genomics to support my PhD studies for the upcoming Fall and Winter quarters.

I presented a poster at the tenth Bay Area Population Genetics meeting based on my work with Dr. Noah Rosenberg on the mathematical properties of the H12 and H2/H1 statistics. The H12 statistic is a haplotype homozygosity statistic used to identify regions of the genome under positive selection, and the H2/H1 statistic is used to distinguish whether the candidate region under selection shows signatures of a hard versus soft sweep. In our paper, Noah and I show that there is an upper bound for H2/H1 as a function of the corresponding H12 value. We apply this upper bound to data and show that it can help facilitate the interpretation of H12 and H2/H1 measured in heterogenous data sets with varying sample sizes and missing data rates.

Here is a copy of my poster:

I attended the Simons Institute workshop titled Computation-Intensive Probabilistic and Statistical Methods for Large-Scale Population Genomics. There was a great lineup of speakers and some opportunities to meet colleagues!

We recently rewrote our paper on identifying soft selective sweeps in Drosophila and posted version 2 on the ArXiv. In this new version, we (i) focus much more on the possibility of complex demographic scenarios generating detected signatures, (ii) carry out extensive ABC computations to estimate the likeliest adaptive theta for all of our peaks, and (iii) investigate the power of our statistics to detect sweeps of varying “softness” either due to varying adaptive theta values or because the sweeping allele starts at varying initial frequency. In the end, our conclusions remain the same: that recent selective sweeps in Drosophila were abundant and primarily soft.

Please feel free to check out version 2 and send us your comments!